Frequency Diffeomorphisms for Efficient Image Registration

@article{Zhang2017FrequencyDF,
  title={Frequency Diffeomorphisms for Efficient Image Registration},
  author={Miaomiao Zhang and Ruizhi Liao and Adrian V. Dalca and Esra Abaci Turk and Jie Luo and Patricia Ellen Grant and Polina Golland},
  journal={Information processing in medical imaging : proceedings of the ... conference},
  year={2017},
  volume={10265},
  pages={
          559-570
        }
}
This paper presents an efficient algorithm for large deformation diffeomorphic metric mapping (LDDMM) with geodesic shooting for image registration. We introduce a novel finite dimensional Fourier representation of diffeomorphic deformations based on the key fact that the high frequency components of a diffeomorphism remain stationary throughout the integration process when computing the deformation associated with smooth velocity fields. We show that manipulating high dimensional… 

Figures from this paper

Fast Diffeomorphic Image Registration via Fourier-Approximated Lie Algebras
TLDR
This paper introduces Fourier-approximated Lie algebras for shooting (FLASH), a fast geodesic shooting algorithm for diffeomorphic image registration that approximate the infinite-dimensional Lie algebra of smooth vector fields with a low-dimensional, bandlimited space.
Efficient Laplace Approximation for Bayesian Registration Uncertainty Quantification
TLDR
A Laplace approximation of Bayesian registration models entirely in a bandlimited space that fully describes the properties of diffeomorphic transformations and dramatically reduces the computational complexity of approximating posterior marginals in the high dimensional imaging space.
Data-Driven Model Order Reduction for Diffeomorphic Image Registration
TLDR
A reduced order model (ROM) is introduced to substantially lower the overall computational cost while maintaining accurate alignment in the context of large deformation diffeomorphic metric mapping (LDDMM) and is demonstrated in neuroimaging applications of pairwise image registration and template estimation for population studies.
Deep Learning for Regularization Prediction in Diffeomorphic Image Registration
TLDR
A predictive model based on deep convolutional neural networks that learns the mapping between pairwise images and the regularization parameter of image registration and demonstrates the effectiveness of the model on both 2D synthetic data and 3D real brain images.
Diffeomorphic Image Registration with Neural Velocity Field
TLDR
This paper proposes a neural representation of continuous velocity field (NeVF) to describe the deformations across two images that has higher flexibility in modeling the complex deformation field and proposes a simple sparse-sampling strategy to reduce the memory consumption.
Unsupervised Learning for Fast Probabilistic Diffeomorphic Registration
TLDR
This paper presents a probabilistic generative model and derive an unsupervised learning-based inference algorithm that makes use of recent developments in convolutional neural networks (CNNs) and results in state of the art accuracy and very fast runtimes, while providing diffeomorphic guarantees and uncertainty estimates.
Band-Limited Stokes Large Deformation Diffeomorphic Metric Mapping
  • Monica Hernandez
  • Computer Science
    IEEE Journal of Biomedical and Health Informatics
  • 2019
TLDR
This paper proposes a novel method for efficient Stokes-LDDMM diffeomorphic registration that poses the constrained variational problem in the space of band-limited vector fields and it is implemented in the GPU.
...
1
2
3
4
...

References

SHOWING 1-10 OF 28 REFERENCES
Finite-Dimensional Lie Algebras for Fast Diffeomorphic Image Registration
TLDR
A novel finite-dimensional Lie algebra structure on the space of bandlimited velocity fields is introduced that can effectively represent initial velocities for diffeomorphic image registration at much lower dimensions than typically used, with little to no loss in registration accuracy.
Diffeomorphic 3D Image Registration via Geodesic Shooting Using an Efficient Adjoint Calculation
TLDR
The key contribution of this work is to provide an accurate estimation of the so-called initial momentum, which is a scalar function encoding the optimal deformation between two images through the Hamiltonian equations of geodesics.
A fast diffeomorphic image registration algorithm
Diffeomorphic demons: Efficient non-parametric image registration
Bayesian Principal Geodesic Analysis in Diffeomorphic Image Registration
TLDR
A generative Bayesian approach for automatic dimensionality reduction of shape variability represented through diffeomorphic mappings is presented, and a latent variable model for principal geodesic analysis (PGA) that provides a probabilistic framework for factor analysis on diffeomorphisms is developed.
Diffeomorphic image matching with left-invariant metrics
The geometric approach to diffeomorphic image registration known as large deformation by diffeomorphic metric mapping (LDDMM) is based on a left action of diffeomorphisms on images, and a
Splines for diffeomorphisms
A Log-Euclidean Framework for Statistics on Diffeomorphisms
TLDR
This article focuses on the computation of statistics of invertible geometrical deformations (i.e., diffeomorphisms), based on the generalization to this type of data of the notion of principal logarithm, which is a simple 3D vector field and well-defined for diffe morphisms close enough to the identity.
Principal Component Based Diffeomorphic Surface Mapping
TLDR
This new algorithm reduces the complexity of the estimation of diffeomorphic transformations by incorporating a shape prior in which a nonlinear diffeomorph shape space is represented by a linear space of initial momenta of Diffeomorphic geodesic flows from a fixed template.
Geodesic Shooting for Computational Anatomy
TLDR
It is shown that this momentum can be also used for describing a deformation of given visual structures, like points, contours or images, and that, it has the same dimension as the described object, as a consequence of the normal momentum constraint the authors introduce.
...
1
2
3
...